Localization Module (RealSense D435i + OpenCV ArUco + IMU Drift-Aware Fusion)

This project provides a localization pipeline using Intel RealSense D435i IMU + RGB, OpenCV ArUco for pose correction, and a drift-aware complementary fusion layer. Designed for a fixed camera orientation facing an ArUco board 20–80 cm away, with cross-shaped motion.

Setup

python -m venv .venv
. .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Calibrations Required

• Camera intrinsics (camera_matrix, dist_coeffs) from RealSense calibration or calib*.json.
• ArUco marker side length in meters (marker_length_m).
• IMU-to-camera extrinsics if using IMU orientation in camera frame.

Run

python main.py

Notes

• ArUco must be facing the camera; lighting should avoid glare.
• The fusion increases reliance on vision when drift is detected.
• For AprilTag support, swap the detector implementation or use apriltag library.